{"id":1093740,"date":"2024-10-15T03:38:35","date_gmt":"2024-10-15T10:38:35","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-blog-post&p=1093740"},"modified":"2024-10-15T03:38:38","modified_gmt":"2024-10-15T10:38:38","slug":"viseval","status":"publish","type":"msr-blog-post","link":"https:\/\/www.microsoft.com\/en-us\/research\/articles\/viseval\/","title":{"rendered":"VIS 2024\u6700\u4f73\u8bba\u6587 | VisEval\uff1a\u63a8\u52a8\u81ea\u7136\u8bed\u8a00\u751f\u6210\u53ef\u89c6\u5316\u7684\u5168\u65b0\u8bc4\u4f30\u6846\u67b6"},"content":{"rendered":"\n
\u7f16\u8005\u6309\uff1a\u968f\u7740\u4eba\u5de5\u667a\u80fd\u6280\u672f\u7684\u5feb\u901f\u53d1\u5c55\uff0c\u6570\u636e\u53ef\u89c6\u5316\u65e5\u6e10\u9ad8\u6548\u3001\u667a\u80fd\u3002\u4f46\u81ea\u52a8\u5316\u751f\u6210\u7684\u56fe\u8868\u662f\u5426\u53ef\u9760\uff0c\u6210\u4e3a\u4e86\u4e9f\u5f85\u89e3\u51b3\u7684\u95ee\u9898\u3002\u5fae\u8f6f\u4e9a\u6d32\u7814\u7a76\u9662\u63d0\u51fa\u4e86 VisEval \u8bc4\u4f30\u6846\u67b6\uff0c\u4e3a\u8fd9\u4e00\u6311\u6218\u63d0\u4f9b\u4e86\u89e3\u51b3\u65b9\u6848\uff0c\u5e76\u56e0\u6b64\u8363\u83b7\u5168\u7403\u53ef\u89c6\u5316\u9886\u57df\u9876\u5c16\u7684\u5b66\u672f\u4f1a\u8bae IEEE VIS 2024 \u7684\u6700\u4f73\u8bba\u6587\u5956\u3002\u901a\u8fc7\u9ad8\u8d28\u91cf\u7684\u6570\u636e\u96c6\u548c\u53ef\u9760\u7684\u81ea\u52a8\u5316\u8bc4\u4f30\u65b9\u6cd5\uff0cVisEval \u4e3a\u6570\u636e\u53ef\u89c6\u5316\u7684\u672a\u6765\u53d1\u5c55\u63d0\u4f9b\u4e86\u575a\u5b9e\u7684\u57fa\u7840\uff0c\u52a9\u529b\u6570\u636e\u53ef\u89c6\u5316\u6280\u672f\u5411\u66f4\u667a\u80fd\u3001\u66f4\u4fbf\u6377\u7684\u65b9\u5411\u53d1\u5c55\u3002<\/p>\n\n\n\n
\u5728\u5982\u4eca\u8fd9\u4e2a\u6570\u636e\u9a71\u52a8\u7684\u65f6\u4ee3\uff0c\u6570\u636e\u53ef\u89c6\u5316\u5df2\u6210\u4e3a\u5c55\u793a\u6570\u636e\u5185\u5728\u4fe1\u606f\u7684\u91cd\u8981\u5de5\u5177\u4e4b\u4e00\u3002\u60f3\u8c61\u4e00\u4e0b\uff0c\u82e5\u53ea\u9700\u4e00\u53e5\u7b80\u5355\u7684\u6307\u4ee4\uff0c\u590d\u6742\u7684\u6570\u636e\u4fbf\u80fd\u201c\u5316\u7e41\u4e3a\u7b80\u201d\uff0c\u5448\u73b0\u4e3a\u76f4\u89c2\u3001\u6613\u4e8e\u7406\u89e3\u7684\u56fe\u8868\uff0c\u90a3\u4e48\u65e2\u53ef\u4ee5\u51cf\u8f7b\u5206\u6790\u6570\u636e\u7684\u538b\u529b\uff0c\u4e5f\u8ba9\u6570\u636e\u7684\u4ea4\u6d41\u4e0e\u4f20\u9012\u53d8\u5f97\u66f4\u4e3a\u8f7b\u677e\u6709\u8da3\u3002\u8fd1\u671f\uff0c\u5f97\u76ca\u4e8e\u5927\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u7684\u7a81\u7834\u6027\u8fdb\u5c55\uff0c\u81ea\u52a8\u5316\u6570\u636e\u53ef\u89c6\u5316\u751f\u6210\u7684\u68a6\u60f3\u9010\u6b65\u5b9e\u73b0\u3002\u7136\u800c\uff0c\u6f5c\u5728\u7684\u95ee\u9898\u4e5f\u968f\u4e4b\u800c\u6765\uff1a\u7531 LLMs \u751f\u6210\u7684\u53ef\u89c6\u5316\u56fe\u8868\uff0c\u771f\u7684\u53ef\u9760\u5417\uff1f\u5b83\u4eec\u662f\u5426\u80fd\u9075\u5faa\u6570\u636e\u53ef\u89c6\u5316\u7684\u6700\u4f73\u5b9e\u8df5\uff1f<\/p>\n\n\n\n
\u4e3a\u4e86\u6709\u6548\u5e94\u5bf9\u8fd9\u4e9b\u6311\u6218\uff0c\u5fae\u8f6f\u4e9a\u6d32\u7814\u7a76\u9662\u63a8\u51fa\u4e86\u4e00\u5957\u5168\u65b0\u7684\u8bc4\u4f30\u6846\u67b6\u2014\u2014VisEval\uff0c\u4e3a\u6570\u636e\u53ef\u89c6\u5316\u751f\u6210\u63d0\u4f9b\u4e86\u66f4\u5168\u9762\u3001\u66f4\u79d1\u5b66\u7684\u8bc4\u4f30\u673a\u5236\u3002VisEval \u4e0d\u4ec5\u6784\u5efa\u4e86\u4e00\u4e2a\u9ad8\u8d28\u91cf\u3001\u8986\u76d6\u5e7f\u6cdb\u7684\u53ef\u89c6\u5316\u6570\u636e\u96c6\uff0c\u8fd8\u901a\u8fc7\u591a\u7ef4\u5ea6\u7684\u8bc4\u4f30\u673a\u5236\uff0c\u4ece\u751f\u6210\u4ee3\u7801\u7684\u6709\u6548\u6027\u5230\u56fe\u8868\u7684\u5951\u5408\u6027\u3001\u53ef\u8bfb\u6027\u8fdb\u884c\u4e86\u5168\u9762\u5ba1\u67e5\u3002\u76f8\u5173\u8bba\u6587\u5df2\u88ab\u5168\u7403\u53ef\u89c6\u5316\u9886\u57df\u9876\u5c16\u7684\u5b66\u672f\u4f1a\u8bae IEEE VIS 2024 \u8bc4\u9009\u4e3a\u6700\u4f73\u8bba\u6587\u3002<\/p>\n\n\n\n \u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/abs\/2407.00981<\/p>\n\n\n\n GitHub \u94fe\u63a5\uff1ahttps:\/\/github.com\/microsoft\/VisEval<\/p>\n\n\n\n \u7528\u5927\u6a21\u578b\u751f\u6210\u53ef\u89c6\u5316\u7684\u8fc7\u7a0b\u901a\u5e38\u5305\u62ec\uff1a\u5c06\u7528\u6237\u7684\u67e5\u8be2\uff08query\uff09\u548c\u6570\u636e\u6574\u5408\u5230\u63d0\u793a\u8bcd\uff08prompt\uff09\u4e2d\uff0c\u7136\u540e\u4f7f\u7528\u8bf8\u5982 Matplotlib \u6216 Seaborn \u7b49\u53ef\u89c6\u5316\u5e93\u751f\u6210\u4ee3\u7801\uff0c\u6700\u7ec8\u5728\u6c99\u76d2\u73af\u5883\u4e2d\u6267\u884c\u8fd9\u4e9b\u4ee3\u7801\u4ee5\u751f\u6210\u56fe\u8868\u3002\u867d\u7136\u8fd9\u4e2a\u8fc7\u7a0b\u542c\u8d77\u6765\u7b80\u5355\uff0c\u4f46\u5b9e\u9645\u4e0a\uff0c\u73b0\u6709\u7684\u5927\u8bed\u8a00\u6a21\u578b\u5728\u751f\u6210\u53ef\u89c6\u5316\u65f6\u4f1a\u9762\u4e34\u8bf8\u591a\u95ee\u9898\u3002<\/p>\n\n\n\n \u5982\u56fe1\u6240\u793a\uff0c\u5f53\u5927\u6a21\u578b\u6839\u636e\u8239\u8236\u6570\u636e\u751f\u6210\u5806\u53e0\u6761\u5f62\u56fe\u65f6\uff0c\u4e0d\u540c\u6a21\u578b\u7684\u8868\u73b0\u5404\u5f02\uff1a\u6709\u7684\u751f\u6210\u7684\u4ee3\u7801\u65e0\u6cd5\u6267\u884c\uff0c\u6709\u7684\u6620\u5c04\u6570\u636e\u51fa\u9519\uff0c\u8fd8\u6709\u7684\u672a\u80fd\u6b63\u786e\u6392\u5e8f\u6216\u56fe\u4f8b\u6446\u653e\u6df7\u4e71\u3002\u8fd9\u4e9b\u95ee\u9898\u53ef\u4ee5\u5f52\u7eb3\u4e3a\u4e09\u5927\u7c7b\uff1a\u6709\u6548\u6027\u3001\u5951\u5408\u6027\u548c\u53ef\u8bfb\u6027\u3002\u5177\u4f53\u800c\u8a00\uff1a<\/p>\n\n\n\n \u53ef\u9760\u3001\u5168\u9762\u7684\u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\u7684\u7f3a\u4e4f\uff0c\u963b\u788d\u4e86\u4eba\u4eec\u5bf9\u5927\u6a21\u578b\u5728\u751f\u6210\u53ef\u89c6\u5316\u65f6\u4e0d\u8db3\u4e4b\u5904\u7684\u8ba4\u8bc6\u3002\u76ee\u524d\uff0c\u53ef\u89c6\u5316\u8bc4\u4f30\u6570\u636e\u96c6\u666e\u904d\u5b58\u5728\u4e00\u4e9b\u5c40\u9650\u6027\uff0c\u9650\u5236\u4e86\u5168\u9762\u6df1\u5165\u7684\u8bc4\u4f30\uff1a\u7f3a\u5c11\u6587\u672c\u67e5\u8be2\u6216\u539f\u59cb\u6570\u636e\u3001\u7f3a\u4e4f\u660e\u786e\u7684\u6807\u51c6\u7b54\u6848\uff0c\u4ee5\u53ca\u8fc7\u4e8e\u4e13\u6ce8\u4e8e\u72ed\u7a84\u9886\u57df\u4e14\u89c4\u6a21\u6709\u9650\u3002<\/p>\n\n\n\n \u6b64\u5916\uff0c\u73b0\u6709\u7684\u8bc4\u4f30\u65b9\u6cd5\u4e5f\u5b58\u5728\u4e0d\u8db3\u3002\u4eba\u5de5\u8bc4\u4f30\u867d\u7136\u88ab\u89c6\u4e3a\u201c\u9ec4\u91d1\u6807\u51c6\u201d\uff0c\u4f46\u5176\u8017\u65f6\u8d39\u529b\uff0c\u96be\u4ee5\u5927\u89c4\u6a21\u63a8\u5e7f\uff1b\u57fa\u4e8e\u89c4\u5219\u7684\u8bc4\u4f30\u80fd\u591f\u68c0\u67e5\u6570\u636e\u7684\u5339\u914d\u60c5\u51b5\uff0c\u4f46\u5e38\u5e38\u5ffd\u89c6\u53ef\u89c6\u5316\u7684\u53ef\u8bfb\u6027\u95ee\u9898\uff1b\u800c\u5229\u7528\u5927\u8bed\u8a00\u6a21\u578b\u8bc4\u4f30\u751f\u6210\u4ee3\u7801\u7684\u65b9\u5f0f\u5c1a\u672a\u7ecf\u8fc7\u5145\u5206\u9a8c\u8bc1\uff0c\u5176\u53ef\u9760\u6027\u5b58\u7591\u3002<\/p>\n\n\n\n VisEval \u7684\u63d0\u51fa\u4e0d\u4ec5\u63d0\u4f9b\u4e86\u4e00\u4e2a\u9ad8\u8d28\u91cf\u7684\u5927\u89c4\u6a21\u6570\u636e\u96c6\uff0c\u8fd8\u5f15\u5165\u4e86\u53ef\u9760\u4e14\u591a\u7ef4\u5ea6\u7684\u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\uff0c\u4ece\u800c\u786e\u4fdd\u5bf9\u751f\u6210\u7684\u53ef\u89c6\u5316\u53ef\u4ee5\u8fdb\u884c\u5168\u9762\u7684\u8bc4\u4f30\u3002<\/p>\n\n\n\n \u5728\u6570\u636e\u96c6\u7684\u6784\u5efa\u8fc7\u7a0b\u4e2d\uff0c\u7814\u7a76\u5458\u4eec\u4e13\u6ce8\u4e8e\u6311\u9009\u65e0\u6b67\u4e49\u3001\u5408\u7406\u4e14\u65e0\u91cd\u590d\u7684\u67e5\u8be2\uff0c\u5e76\u4e3a\u6bcf\u4e2a\u67e5\u8be2\u63d0\u4f9b\u4e86\u51c6\u786e\u7684\u6807\u51c6\u7b54\u6848\uff0c\u6765\u4fdd\u8bc1\u6570\u636e\u7684\u53ef\u9760\u6027\u548c\u6709\u6548\u6027\u3002\u4e3a\u4e86\u589e\u5f3a\u6570\u636e\u96c6\u7684\u9c81\u68d2\u6027\uff0c\u7814\u7a76\u5458\u4eec\u5728\u6570\u636e\u96c6\u4e2d\u6db5\u76d6\u4e86\u591a\u4e2a\u9886\u57df\u548c\u56fe\u8868\u7c7b\u578b\uff0c\u540c\u65f6\u6392\u9664\u4e86\u8fc7\u4e8e\u7b80\u5355\u7684\u67e5\u8be2\uff0c\u4ee5\u786e\u4fdd\u751f\u6210\u7684\u53ef\u89c6\u5316\u5177\u6709\u4e00\u5b9a\u7684\u590d\u6742\u6027\u3002\u57fa\u4e8e nvBench \u6570\u636e\u96c6\uff0c\u7814\u7a76\u5458\u4eec\u7ed3\u5408\u5927\u8bed\u8a00\u6a21\u578b\u548c\u4eba\u5de5\u4e13\u5bb6\u7684\u7b5b\u9009\uff0c\u7cbe\u5fc3\u6311\u9009\u51fa\u4e86\u9ad8\u8d28\u91cf\u7684\u67e5\u8be2\u3002\u8fd9\u4e00\u8fc7\u7a0b\u4e0d\u4ec5\u517c\u987e\u4e86\u6570\u636e\u7684\u8d28\u91cf\uff0c\u8fd8\u6709\u6548\u51cf\u5c11\u4e86\u4eba\u529b\u5de5\u4f5c\u8d1f\u62c5\u3002\u6b64\u5916\uff0c\u7814\u7a76\u5458\u4eec\u901a\u8fc7\u5143\u4fe1\u606f\uff08meta information\uff09\u4e3a\u6bcf\u4e2a\u67e5\u8be2\u6ce8\u91ca\u4e86\u6240\u6709\u53ef\u63a5\u53d7\u7684\u6807\u51c6\u7b54\u6848\uff0c\u5e76\u8fdb\u884c\u4e86\u6570\u636e\u96c6\u7684\u91cd\u65b0\u5e73\u8861\u3002\u8fd9\u4e00\u7cfb\u5217\u63aa\u65bd\u90fd\u786e\u4fdd\u4e86 VisEval \u6570\u636e\u96c6\u7684\u5168\u9762\u6027\u548c\u5b9e\u7528\u6027\uff0c\u4f7f\u5176\u5728\u81ea\u52a8\u5316\u6570\u636e\u53ef\u89c6\u5316\u7684\u8bc4\u4f30\u4e2d\u53ef\u4ee5\u53d1\u6325\u66f4\u5927\u7684\u4f5c\u7528\u3002\u6700\u7ec8\uff0cVisEval \u7684\u6570\u636e\u96c6\u4e2d\u5305\u542b\u4e867\u79cd\u56fe\u8868\u7c7b\u578b\uff0c\u8d85\u8fc71,000\u4e2a\u53ef\u89c6\u5316\u56fe\u8868\u3002<\/p>\n\n\n\n \u6570\u636e\u96c6\u6784\u5efa\u5b8c\u6210\u540e\uff0c\u7814\u7a76\u5458\u4eec\u53c8\u5f00\u53d1\u4e86\u4e00\u4e2a\u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\u3002\u8be5\u6846\u67b6\u5982\u56fe3\u6240\u793a\uff0c\u5206\u4e3a\u4e09\u4e2a\u6a21\u5757\uff0c\u5206\u522b\u5bf9\u751f\u6210\u4ee3\u7801\u7684\u6709\u6548\u6027\uff08validity\uff09\u3001\u5951\u5408\u6027\uff08legality\uff09\u548c\u53ef\u8bfb\u6027\uff08readability\uff09\u8fdb\u884c\u8bc4\u4f30\u3002\u4e3a\u4e86\u4fdd\u8bc1\u8bc4\u4f30\u6846\u67b6\u7684\u53ef\u9760\u6027\uff0c\u7814\u7a76\u5458\u4eec\u8fdb\u884c\u4e86\u7ec6\u81f4\u7684\u5ba1\u67e5\uff0c\u91cd\u70b9\u5173\u6ce8\u6570\u636e\u96c6\u548c\u8bc4\u4f30\u6846\u67b6\u7684\u5404\u4e2a\u65b9\u9762\u3002\u6b64\u5916\uff0c\u7814\u7a76\u5458\u4eec\u901a\u8fc7\u8bbe\u8ba1\u6d4b\u8bd5\u7528\u4f8b\u548c\u8fdb\u884c\u5b9a\u91cf\u8bc4\u4f30\uff0c\u6765\u9a8c\u8bc1\u8bc4\u4f30\u7684\u8d28\u91cf\u3002\u901a\u8fc7\u8fd9\u4e9b\u6b65\u9aa4\uff0c\u7814\u7a76\u5458\u4eec\u80fd\u591f\u786e\u4fdd\u8bc4\u4f30\u6846\u67b6\u53ef\u4ee5\u51c6\u786e\u3001\u5168\u9762\u5730\u5bf9\u751f\u6210\u7684\u53ef\u89c6\u5316\u8fdb\u884c\u6709\u6548\u8bc4\u4f30\u3002<\/p>\n\n\n\n \u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\u5177\u4f53\u5982\u4e0b\uff1a<\/p>\n\n\n\n \u4ee3\u7801\u751f\u6210\u540e\uff0c\u6846\u67b6\u4f1a\u6267\u884c\u4ee3\u7801\uff0c\u5e76\u68c0\u67e5\u662f\u5426\u6210\u529f\u751f\u6210\u4e86\u53ef\u89c6\u5316\uff0c\u4ee5\u786e\u4fdd\u4ee3\u7801\u7684\u6709\u6548\u6027\u3002<\/p>\n\n\n\n \u4e00\u65e6\u4ee3\u7801\u901a\u8fc7\u4e86\u6709\u6548\u6027\u68c0\u67e5\uff0c\u5951\u5408\u6027\u68c0\u67e5\u6a21\u5757\u4f1a\u63d0\u53d6\u56fe\u8868\u7c7b\u578b\u3001\u6570\u636e\u7b49\u4fe1\u606f\uff0c\u5e76\u6839\u636e\u6570\u636e\u96c6\u4e2d\u6ce8\u91ca\u7684\u5143\u4fe1\u606f\u8bc4\u4f30\u56fe\u8868\u7684\u5951\u5408\u6027\u3002\u8fd9\u4e00\u6b65\u662f\u4e3a\u4e86\u786e\u4fdd\u751f\u6210\u7684\u56fe\u8868\u80fd\u591f\u7b26\u5408\u7528\u6237\u67e5\u8be2\u7684\u8981\u6c42\uff0c\u5e76\u4e14\u6570\u636e\u6620\u5c04\u5408\u7406\u3002<\/p>\n\n\n\n \u53ef\u8bfb\u6027\u8bc4\u4f30\u662f\u6846\u67b6\u4e2d\u6700\u5177\u6311\u6218\u6027\u7684\u90e8\u5206\u3002\u8bc4\u4f30\u53ef\u8bfb\u6027\u9700\u8981\u8003\u8651\u591a\u65b9\u9762\u56e0\u7d20\uff0c\u4f8b\u5982\u5e03\u5c40\u3001\u6bd4\u4f8b\u548c\u989c\u8272\u7b49\u3002\u4e3a\u6b64\uff0c\u7814\u7a76\u5458\u4eec\u501f\u52a9\u4e86 GPT-4V(ISION) \u7684\u5f3a\u5927\u80fd\u529b\uff0c\u5e76\u5c06\u53ef\u8bfb\u6027\u8bc4\u4f30\u4efb\u52a1\u5206\u89e3\u4e3a\u591a\u4e2a\u66f4\u53ef\u63a7\u7684\u5b50\u95ee\u9898\u3002\u5148\u5206\u522b\u5bf9\u5e03\u5c40\u3001\u6bd4\u4f8b\u5c3a\u548c\u523b\u5ea6\u8fdb\u884c\u7ec6\u5316\u8bc4\u4f30\uff0c\u518d\u5c06\u8bc4\u4f30\u7ed3\u679c\u6c47\u603b\uff0c\u6574\u4f53\u7ed9\u51fa\u53ef\u8bfb\u6027\u7684\u6253\u5206\u3002\u7814\u7a76\u5458\u4eec\u7684\u5b9a\u91cf\u5b9e\u9a8c\u8868\u660e\uff0cVisEval \u7684\u53ef\u8bfb\u6027\u8bc4\u4f30\u7ed3\u679c\u4e0e\u4eba\u7c7b\u504f\u597d\u9ad8\u5ea6\u4e00\u81f4\u3002<\/p>\n\n\n\n \u57fa\u4e8e VisEval \u6846\u67b6\uff0c\u7814\u7a76\u5458\u4eec\u5bf9\u591a\u79cd\u6a21\u578b\u7684\u53ef\u89c6\u5316\u751f\u6210\u80fd\u529b\u8fdb\u884c\u4e86\u7cfb\u7edf\u8bc4\u4f30\u3002<\/p>\n\n\n\n \u7b2c\u4e00\u9879\u5b9e\u9a8c\u6d4b\u8bd5\u4e86\u4e0d\u540c\u6a21\u578b\u5728\u4f7f\u7528 Matplotlib \u548c Seaborn \u5e93\u65f6\u7684\u8868\u73b0\u3002\u5982\u88681\u6240\u793a\uff0cVisEval \u63ed\u793a\u4e86\u4e0d\u540c\u6a21\u578b\u5728\u751f\u6210\u53ef\u89c6\u5316\u65b9\u9762\u7684\u663e\u8457\u5dee\u5f02\u3002\u4ee5 GPT-4 \u4e3a\u4f8b\uff0c\u5728\u4f7f\u7528 Matplotlib \u5e93\u751f\u6210\u53ef\u89c6\u5316\u65f6\uff0c\u5176\u8d28\u91cf\u5f97\u5206\u4e3a2.89\uff08\u6ee1\u5206\u4e3a5\uff09\uff0c\u867d\u7136\u8868\u73b0\u5c1a\u53ef\uff0c\u4f46\u4f9d\u7136\u6709\u6539\u8fdb\u7684\u7a7a\u95f4\u3002\u76f8\u5bf9\u800c\u8a00\uff0c\u5728\u4f7f\u7528 Seaborn \u5e93\u65f6\uff0c\u5c3d\u7ba1\u5176\u4ee3\u7801\u901a\u5e38\u6bd4 Matplotlib \u66f4\u4e3a\u7b80\u6d01\uff0c\u4f46\u6240\u6709\u6a21\u578b\u7684\u5f97\u5206\u5747\u8f83\u4f4e\uff0c\u8fd9\u4e00\u7ed3\u679c\u4ee4\u4eba\u610f\u5916\u3002<\/p>\n\n\n\n \u901a\u8fc7\u6df1\u5165\u5206\u6790\uff0c\u7814\u7a76\u5458\u4eec\u53d1\u73b0\u5927\u8bed\u8a00\u6a21\u578b\u5728\u53ef\u89c6\u5316\u751f\u6210\u7684\u591a\u4e2a\u9636\u6bb5\u90fd\u5bb9\u6613\u51fa\u73b0\u9519\u8bef\uff0c\u5982\u56fe4\u6240\u793a\u3002\u8fd9\u4e9b\u9636\u6bb5\u5305\u62ec\u4ee3\u7801\u7f16\u5199\u3001\u6570\u636e\u8f6c\u6362\u3001\u53ef\u89c6\u5316\u8f6c\u6362\u4ee5\u53ca\u6392\u5e8f\u7b49\u65b9\u9762\u3002\u9664\u4e86\u51c6\u786e\u6027\u4e4b\u5916\uff0c\u6a21\u578b\u751f\u6210\u7684\u56fe\u8868\u5728\u53ef\u8bfb\u6027\u65b9\u9762\u4e5f\u9891\u7e41\u51fa\u73b0\u95ee\u9898\uff0c\u800c\u8fd9\u4e00\u70b9\u5728\u4ee5\u5f80\u7684\u7814\u7a76\u4e2d\u5e38\u5e38\u88ab\u5ffd\u89c6\u3002<\/p>\n\n\n\n \u5728\u7b2c\u4e8c\u9879\u5b9e\u9a8c\u4e2d\uff0c\u7814\u7a76\u5458\u4eec\u63a2\u8ba8\u4e86\u4e0d\u540c\u63d0\u793a\u8bcd\u8bbe\u8ba1\u5bf9\u6a21\u578b\u6027\u80fd\u7684\u5f71\u54cd\u3002\u5176\u4e2d\u7279\u522b\u5206\u6790\u4e86\u4e09\u79cd\u57fa\u4e8e\u5927\u8bed\u8a00\u6a21\u578b\u7684\u53ef\u89c6\u5316\u751f\u6210\u65b9\u6cd5\uff1aLIDA\u3001Chat2VIS \u548c\u7814\u7a76\u56e2\u961f\u63d0\u51fa\u7684 CoML4VIS\u3002\u5b9e\u9a8c\u7ed3\u679c\u8868\u660e\uff0c\u63d0\u793a\u8bcd\u8bbe\u8ba1\u5bf9\u6a21\u578b\u7684\u8868\u73b0\u5177\u6709\u663e\u8457\u5f71\u54cd\u3002\u7531\u4e8e\u6ce8\u610f\u5230\u8fd9\u4e09\u79cd\u65b9\u6cd5\u91c7\u7528\u4e86\u4e0d\u540c\u7684\u8868\u683c\u683c\u5f0f\uff0c\u6240\u4ee5\u7814\u7a76\u5458\u4eec\u8fd8\u8fdb\u884c\u4e86\u989d\u5916\u7684\u5b9e\u9a8c\u3002\u5b9e\u9a8c\u4e2d\u7814\u7a76\u5458\u4eec\u4fdd\u6301 CoML4VIS \u4e2d\u5176\u4ed6\u6761\u4ef6\u4e0d\u53d8\uff0c\u4ec5\u66f4\u6539\u8868\u683c\u683c\u5f0f\u3002\u5982\u56fe5\u6240\u793a\uff0c\u4e0d\u540c\u7684\u5927\u8bed\u8a00\u6a21\u578b\u5c55\u73b0\u51fa\u4e86\u5bf9\u4e0d\u540c\u8868\u683c\u683c\u5f0f\u7684\u504f\u597d\u3002\u8fd9\u4e00\u53d1\u73b0\u63d0\u793a\u7814\u7a76\u5458\u4eec\uff0c\u53ef\u80fd\u9700\u8981\u9488\u5bf9\u4e0d\u540c\u6a21\u578b\u8bbe\u8ba1\u4e0d\u540c\u7684\u63d0\u793a\u8bcd\uff0c\u4ee5\u4f18\u5316\u5176\u8868\u73b0\u3002<\/p>\n\n\n\n \u5728\u7b2c\u4e09\u9879\u5b9e\u9a8c\u4e2d\uff0c\u7814\u7a76\u5458\u4eec\u6d4b\u8bd5\u4e86\u65e0\u7528\u6570\u636e\u8868\u5bf9\u751f\u6210\u53ef\u89c6\u5316\u6027\u80fd\u7684\u5f71\u54cd\u3002\u5982\u88682\u6240\u793a\uff0c\u5f53\u7ed9\u5927\u6a21\u578b\u7684\u63d0\u793a\u8bcd\u4e2d\u5305\u542b\u4e24\u5f20\u65e0\u7528\u7684\u6570\u636e\u8868\u65f6\uff0c\u6240\u6709\u6a21\u578b\u7684\u6027\u80fd\u5747\u663e\u8457\u4e0b\u964d\u3002\u8fd9\u4e00\u7ed3\u679c\u8868\u660e\uff0c\u6a21\u578b\u5728\u5904\u7406\u590d\u6742\u8f93\u5165\u65f6\u5bb9\u6613\u53d7\u5230\u5e72\u6270\uff0c\u5f3a\u8c03\u4e86\u5728\u9009\u62e9\u6570\u636e\u65f6\u786e\u4fdd\u5176\u76f8\u5173\u6027\u7684\u91cd\u8981\u4f5c\u7528\u3002<\/p>\n\n\n\n \u4f5c\u4e3a\u4e00\u79cd\u5168\u65b0\u7684\u53ef\u89c6\u5316\u751f\u6210\u8bc4\u4f30\u6846\u67b6\uff0cVisEval \u901a\u8fc7\u9ad8\u8d28\u91cf\u7684\u6570\u636e\u96c6\u548c\u53ef\u9760\u7684\u81ea\u52a8\u5316\u8bc4\u4f30\u65b9\u6cd5\uff0c\u586b\u8865\u4e86\u73b0\u6709\u8bc4\u4f30\u4f53\u7cfb\u7684\u7a7a\u767d\uff0c\u4e3a\u672a\u6765\u7684\u53ef\u89c6\u5316\u751f\u6210\u7814\u7a76\u63d0\u4f9b\u4e86\u91cd\u8981\u7684\u53c2\u8003\u6807\u51c6\u3002\u5fae\u8f6f\u4e9a\u6d32\u7814\u7a76\u9662\u7684\u7814\u7a76\u5458\u4eec\u671f\u5f85\uff0c\u968f\u7740\u6570\u636e\u53ef\u89c6\u5316\u6280\u672f\u53d8\u5f97\u66f4\u52a0\u667a\u80fd\u548c\u4fbf\u6377\uff0c\u5404\u884c\u5404\u4e1a\u7684\u6570\u636e\u9a71\u52a8\u51b3\u7b56\u5c06\u5f97\u5230\u66f4\u6709\u529b\u7684\u652f\u6301\u3002<\/p>\n","protected":false},"excerpt":{"rendered":" \u7f16\u8005\u6309\uff1a\u968f\u7740\u4eba\u5de5\u667a\u80fd\u6280\u672f\u7684\u5feb\u901f\u53d1\u5c55\uff0c\u6570\u636e\u53ef\u89c6\u5316\u65e5\u6e10\u9ad8\u6548\u3001\u667a\u80fd\u3002\u4f46\u81ea\u52a8\u5316\u751f\u6210\u7684\u56fe\u8868\u662f\u5426\u53ef\u9760\uff0c\u6210\u4e3a\u4e86\u4e9f\u5f85\u89e3\u51b3\u7684\u95ee\u9898\u3002\u5fae\u8f6f\u4e9a\u6d32\u7814\u7a76\u9662\u63d0\u51fa\u4e86 VisEval \u8bc4\u4f30\u6846\u67b6\uff0c\u4e3a\u8fd9\u4e00\u6311\u6218\u63d0\u4f9b\u4e86\u89e3\u51b3\u65b9\u6848\uff0c\u5e76\u56e0\u6b64\u8363\u83b7\u5168\u7403\u53ef\u89c6\u5316\u9886\u57df\u9876\u5c16\u7684\u5b66\u672f\u4f1a\u8bae IEEE VIS 2024 \u7684\u6700\u4f73\u8bba\u6587\u5956\u3002\u901a\u8fc7\u9ad8\u8d28\u91cf\u7684\u6570\u636e\u96c6\u548c\u53ef\u9760\u7684\u81ea\u52a8\u5316\u8bc4\u4f30\u65b9\u6cd5\uff0cVisEval \u4e3a\u6570\u636e\u53ef\u89c6\u5316\u7684\u672a\u6765\u53d1\u5c55\u63d0\u4f9b\u4e86\u575a\u5b9e\u7684\u57fa\u7840\uff0c\u52a9\u529b\u6570\u636e\u53ef\u89c6\u5316\u6280\u672f\u5411\u66f4\u667a\u80fd\u3001\u66f4\u4fbf\u6377\u7684\u65b9\u5411\u53d1\u5c55\u3002 \u5728\u5982\u4eca\u8fd9\u4e2a\u6570\u636e\u9a71\u52a8\u7684\u65f6\u4ee3\uff0c\u6570\u636e\u53ef\u89c6\u5316\u5df2\u6210\u4e3a\u5c55\u793a\u6570\u636e\u5185\u5728\u4fe1\u606f\u7684\u91cd\u8981\u5de5\u5177\u4e4b\u4e00\u3002\u60f3\u8c61\u4e00\u4e0b\uff0c\u82e5\u53ea\u9700\u4e00\u53e5\u7b80\u5355\u7684\u6307\u4ee4\uff0c\u590d\u6742\u7684\u6570\u636e\u4fbf\u80fd\u201c\u5316\u7e41\u4e3a\u7b80\u201d\uff0c\u5448\u73b0\u4e3a\u76f4\u89c2\u3001\u6613\u4e8e\u7406\u89e3\u7684\u56fe\u8868\uff0c\u90a3\u4e48\u65e2\u53ef\u4ee5\u51cf\u8f7b\u5206\u6790\u6570\u636e\u7684\u538b\u529b\uff0c\u4e5f\u8ba9\u6570\u636e\u7684\u4ea4\u6d41\u4e0e\u4f20\u9012\u53d8\u5f97\u66f4\u4e3a\u8f7b\u677e\u6709\u8da3\u3002\u8fd1\u671f\uff0c\u5f97\u76ca\u4e8e\u5927\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u7684\u7a81\u7834\u6027\u8fdb\u5c55\uff0c\u81ea\u52a8\u5316\u6570\u636e\u53ef\u89c6\u5316\u751f\u6210\u7684\u68a6\u60f3\u9010\u6b65\u5b9e\u73b0\u3002\u7136\u800c\uff0c\u6f5c\u5728\u7684\u95ee\u9898\u4e5f\u968f\u4e4b\u800c\u6765\uff1a\u7531 LLMs \u751f\u6210\u7684\u53ef\u89c6\u5316\u56fe\u8868\uff0c\u771f\u7684\u53ef\u9760\u5417\uff1f\u5b83\u4eec\u662f\u5426\u80fd\u9075\u5faa\u6570\u636e\u53ef\u89c6\u5316\u7684\u6700\u4f73\u5b9e\u8df5\uff1f \u4e3a\u4e86\u6709\u6548\u5e94\u5bf9\u8fd9\u4e9b\u6311\u6218\uff0c\u5fae\u8f6f\u4e9a\u6d32\u7814\u7a76\u9662\u63a8\u51fa\u4e86\u4e00\u5957\u5168\u65b0\u7684\u8bc4\u4f30\u6846\u67b6\u2014\u2014VisEval\uff0c\u4e3a\u6570\u636e\u53ef\u89c6\u5316\u751f\u6210\u63d0\u4f9b\u4e86\u66f4\u5168\u9762\u3001\u66f4\u79d1\u5b66\u7684\u8bc4\u4f30\u673a\u5236\u3002VisEval \u4e0d\u4ec5\u6784\u5efa\u4e86\u4e00\u4e2a\u9ad8\u8d28\u91cf\u3001\u8986\u76d6\u5e7f\u6cdb\u7684\u53ef\u89c6\u5316\u6570\u636e\u96c6\uff0c\u8fd8\u901a\u8fc7\u591a\u7ef4\u5ea6\u7684\u8bc4\u4f30\u673a\u5236\uff0c\u4ece\u751f\u6210\u4ee3\u7801\u7684\u6709\u6548\u6027\u5230\u56fe\u8868\u7684\u5951\u5408\u6027\u3001\u53ef\u8bfb\u6027\u8fdb\u884c\u4e86\u5168\u9762\u5ba1\u67e5\u3002\u76f8\u5173\u8bba\u6587\u5df2\u88ab\u5168\u7403\u53ef\u89c6\u5316\u9886\u57df\u9876\u5c16\u7684\u5b66\u672f\u4f1a\u8bae IEEE VIS 2024 \u8bc4\u9009\u4e3a\u6700\u4f73\u8bba\u6587\u3002 \u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/abs\/2407.00981 GitHub \u94fe\u63a5\uff1ahttps:\/\/github.com\/microsoft\/VisEval \u7528\u5927\u6a21\u578b\u751f\u6210\u53ef\u89c6\u5316\u7684\u8fc7\u7a0b\u901a\u5e38\u5305\u62ec\uff1a\u5c06\u7528\u6237\u7684\u67e5\u8be2\uff08query\uff09\u548c\u6570\u636e\u6574\u5408\u5230\u63d0\u793a\u8bcd\uff08prompt\uff09\u4e2d\uff0c\u7136\u540e\u4f7f\u7528\u8bf8\u5982 Matplotlib \u6216 Seaborn \u7b49\u53ef\u89c6\u5316\u5e93\u751f\u6210\u4ee3\u7801\uff0c\u6700\u7ec8\u5728\u6c99\u76d2\u73af\u5883\u4e2d\u6267\u884c\u8fd9\u4e9b\u4ee3\u7801\u4ee5\u751f\u6210\u56fe\u8868\u3002\u867d\u7136\u8fd9\u4e2a\u8fc7\u7a0b\u542c\u8d77\u6765\u7b80\u5355\uff0c\u4f46\u5b9e\u9645\u4e0a\uff0c\u73b0\u6709\u7684\u5927\u8bed\u8a00\u6a21\u578b\u5728\u751f\u6210\u53ef\u89c6\u5316\u65f6\u4f1a\u9762\u4e34\u8bf8\u591a\u95ee\u9898\u3002 \u5982\u56fe1\u6240\u793a\uff0c\u5f53\u5927\u6a21\u578b\u6839\u636e\u8239\u8236\u6570\u636e\u751f\u6210\u5806\u53e0\u6761\u5f62\u56fe\u65f6\uff0c\u4e0d\u540c\u6a21\u578b\u7684\u8868\u73b0\u5404\u5f02\uff1a\u6709\u7684\u751f\u6210\u7684\u4ee3\u7801\u65e0\u6cd5\u6267\u884c\uff0c\u6709\u7684\u6620\u5c04\u6570\u636e\u51fa\u9519\uff0c\u8fd8\u6709\u7684\u672a\u80fd\u6b63\u786e\u6392\u5e8f\u6216\u56fe\u4f8b\u6446\u653e\u6df7\u4e71\u3002\u8fd9\u4e9b\u95ee\u9898\u53ef\u4ee5\u5f52\u7eb3\u4e3a\u4e09\u5927\u7c7b\uff1a\u6709\u6548\u6027\u3001\u5951\u5408\u6027\u548c\u53ef\u8bfb\u6027\u3002\u5177\u4f53\u800c\u8a00\uff1a \u53ef\u9760\u3001\u5168\u9762\u7684\u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\u7684\u7f3a\u4e4f\uff0c\u963b\u788d\u4e86\u4eba\u4eec\u5bf9\u5927\u6a21\u578b\u5728\u751f\u6210\u53ef\u89c6\u5316\u65f6\u4e0d\u8db3\u4e4b\u5904\u7684\u8ba4\u8bc6\u3002\u76ee\u524d\uff0c\u53ef\u89c6\u5316\u8bc4\u4f30\u6570\u636e\u96c6\u666e\u904d\u5b58\u5728\u4e00\u4e9b\u5c40\u9650\u6027\uff0c\u9650\u5236\u4e86\u5168\u9762\u6df1\u5165\u7684\u8bc4\u4f30\uff1a\u7f3a\u5c11\u6587\u672c\u67e5\u8be2\u6216\u539f\u59cb\u6570\u636e\u3001\u7f3a\u4e4f\u660e\u786e\u7684\u6807\u51c6\u7b54\u6848\uff0c\u4ee5\u53ca\u8fc7\u4e8e\u4e13\u6ce8\u4e8e\u72ed\u7a84\u9886\u57df\u4e14\u89c4\u6a21\u6709\u9650\u3002 \u6b64\u5916\uff0c\u73b0\u6709\u7684\u8bc4\u4f30\u65b9\u6cd5\u4e5f\u5b58\u5728\u4e0d\u8db3\u3002\u4eba\u5de5\u8bc4\u4f30\u867d\u7136\u88ab\u89c6\u4e3a\u201c\u9ec4\u91d1\u6807\u51c6\u201d\uff0c\u4f46\u5176\u8017\u65f6\u8d39\u529b\uff0c\u96be\u4ee5\u5927\u89c4\u6a21\u63a8\u5e7f\uff1b\u57fa\u4e8e\u89c4\u5219\u7684\u8bc4\u4f30\u80fd\u591f\u68c0\u67e5\u6570\u636e\u7684\u5339\u914d\u60c5\u51b5\uff0c\u4f46\u5e38\u5e38\u5ffd\u89c6\u53ef\u89c6\u5316\u7684\u53ef\u8bfb\u6027\u95ee\u9898\uff1b\u800c\u5229\u7528\u5927\u8bed\u8a00\u6a21\u578b\u8bc4\u4f30\u751f\u6210\u4ee3\u7801\u7684\u65b9\u5f0f\u5c1a\u672a\u7ecf\u8fc7\u5145\u5206\u9a8c\u8bc1\uff0c\u5176\u53ef\u9760\u6027\u5b58\u7591\u3002 VisEval \u7684\u63d0\u51fa\u4e0d\u4ec5\u63d0\u4f9b\u4e86\u4e00\u4e2a\u9ad8\u8d28\u91cf\u7684\u5927\u89c4\u6a21\u6570\u636e\u96c6\uff0c\u8fd8\u5f15\u5165\u4e86\u53ef\u9760\u4e14\u591a\u7ef4\u5ea6\u7684\u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\uff0c\u4ece\u800c\u786e\u4fdd\u5bf9\u751f\u6210\u7684\u53ef\u89c6\u5316\u53ef\u4ee5\u8fdb\u884c\u5168\u9762\u7684\u8bc4\u4f30\u3002 \u5728\u6570\u636e\u96c6\u7684\u6784\u5efa\u8fc7\u7a0b\u4e2d\uff0c\u7814\u7a76\u5458\u4eec\u4e13\u6ce8\u4e8e\u6311\u9009\u65e0\u6b67\u4e49\u3001\u5408\u7406\u4e14\u65e0\u91cd\u590d\u7684\u67e5\u8be2\uff0c\u5e76\u4e3a\u6bcf\u4e2a\u67e5\u8be2\u63d0\u4f9b\u4e86\u51c6\u786e\u7684\u6807\u51c6\u7b54\u6848\uff0c\u6765\u4fdd\u8bc1\u6570\u636e\u7684\u53ef\u9760\u6027\u548c\u6709\u6548\u6027\u3002\u4e3a\u4e86\u589e\u5f3a\u6570\u636e\u96c6\u7684\u9c81\u68d2\u6027\uff0c\u7814\u7a76\u5458\u4eec\u5728\u6570\u636e\u96c6\u4e2d\u6db5\u76d6\u4e86\u591a\u4e2a\u9886\u57df\u548c\u56fe\u8868\u7c7b\u578b\uff0c\u540c\u65f6\u6392\u9664\u4e86\u8fc7\u4e8e\u7b80\u5355\u7684\u67e5\u8be2\uff0c\u4ee5\u786e\u4fdd\u751f\u6210\u7684\u53ef\u89c6\u5316\u5177\u6709\u4e00\u5b9a\u7684\u590d\u6742\u6027\u3002\u57fa\u4e8e nvBench \u6570\u636e\u96c6\uff0c\u7814\u7a76\u5458\u4eec\u7ed3\u5408\u5927\u8bed\u8a00\u6a21\u578b\u548c\u4eba\u5de5\u4e13\u5bb6\u7684\u7b5b\u9009\uff0c\u7cbe\u5fc3\u6311\u9009\u51fa\u4e86\u9ad8\u8d28\u91cf\u7684\u67e5\u8be2\u3002\u8fd9\u4e00\u8fc7\u7a0b\u4e0d\u4ec5\u517c\u987e\u4e86\u6570\u636e\u7684\u8d28\u91cf\uff0c\u8fd8\u6709\u6548\u51cf\u5c11\u4e86\u4eba\u529b\u5de5\u4f5c\u8d1f\u62c5\u3002\u6b64\u5916\uff0c\u7814\u7a76\u5458\u4eec\u901a\u8fc7\u5143\u4fe1\u606f\uff08meta information\uff09\u4e3a\u6bcf\u4e2a\u67e5\u8be2\u6ce8\u91ca\u4e86\u6240\u6709\u53ef\u63a5\u53d7\u7684\u6807\u51c6\u7b54\u6848\uff0c\u5e76\u8fdb\u884c\u4e86\u6570\u636e\u96c6\u7684\u91cd\u65b0\u5e73\u8861\u3002\u8fd9\u4e00\u7cfb\u5217\u63aa\u65bd\u90fd\u786e\u4fdd\u4e86 VisEval \u6570\u636e\u96c6\u7684\u5168\u9762\u6027\u548c\u5b9e\u7528\u6027\uff0c\u4f7f\u5176\u5728\u81ea\u52a8\u5316\u6570\u636e\u53ef\u89c6\u5316\u7684\u8bc4\u4f30\u4e2d\u53ef\u4ee5\u53d1\u6325\u66f4\u5927\u7684\u4f5c\u7528\u3002\u6700\u7ec8\uff0cVisEval \u7684\u6570\u636e\u96c6\u4e2d\u5305\u542b\u4e867\u79cd\u56fe\u8868\u7c7b\u578b\uff0c\u8d85\u8fc71,000\u4e2a\u53ef\u89c6\u5316\u56fe\u8868\u3002 \u6570\u636e\u96c6\u6784\u5efa\u5b8c\u6210\u540e\uff0c\u7814\u7a76\u5458\u4eec\u53c8\u5f00\u53d1\u4e86\u4e00\u4e2a\u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\u3002\u8be5\u6846\u67b6\u5982\u56fe3\u6240\u793a\uff0c\u5206\u4e3a\u4e09\u4e2a\u6a21\u5757\uff0c\u5206\u522b\u5bf9\u751f\u6210\u4ee3\u7801\u7684\u6709\u6548\u6027\uff08validity\uff09\u3001\u5951\u5408\u6027\uff08legality\uff09\u548c\u53ef\u8bfb\u6027\uff08readability\uff09\u8fdb\u884c\u8bc4\u4f30\u3002\u4e3a\u4e86\u4fdd\u8bc1\u8bc4\u4f30\u6846\u67b6\u7684\u53ef\u9760\u6027\uff0c\u7814\u7a76\u5458\u4eec\u8fdb\u884c\u4e86\u7ec6\u81f4\u7684\u5ba1\u67e5\uff0c\u91cd\u70b9\u5173\u6ce8\u6570\u636e\u96c6\u548c\u8bc4\u4f30\u6846\u67b6\u7684\u5404\u4e2a\u65b9\u9762\u3002\u6b64\u5916\uff0c\u7814\u7a76\u5458\u4eec\u901a\u8fc7\u8bbe\u8ba1\u6d4b\u8bd5\u7528\u4f8b\u548c\u8fdb\u884c\u5b9a\u91cf\u8bc4\u4f30\uff0c\u6765\u9a8c\u8bc1\u8bc4\u4f30\u7684\u8d28\u91cf\u3002\u901a\u8fc7\u8fd9\u4e9b\u6b65\u9aa4\uff0c\u7814\u7a76\u5458\u4eec\u80fd\u591f\u786e\u4fdd\u8bc4\u4f30\u6846\u67b6\u53ef\u4ee5\u51c6\u786e\u3001\u5168\u9762\u5730\u5bf9\u751f\u6210\u7684\u53ef\u89c6\u5316\u8fdb\u884c\u6709\u6548\u8bc4\u4f30\u3002 \u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6\u5177\u4f53\u5982\u4e0b\uff1a \u4ee3\u7801\u751f\u6210\u540e\uff0c\u6846\u67b6\u4f1a\u6267\u884c\u4ee3\u7801\uff0c\u5e76\u68c0\u67e5\u662f\u5426\u6210\u529f\u751f\u6210\u4e86\u53ef\u89c6\u5316\uff0c\u4ee5\u786e\u4fdd\u4ee3\u7801\u7684\u6709\u6548\u6027\u3002 \u4e00\u65e6\u4ee3\u7801\u901a\u8fc7\u4e86\u6709\u6548\u6027\u68c0\u67e5\uff0c\u5951\u5408\u6027\u68c0\u67e5\u6a21\u5757\u4f1a\u63d0\u53d6\u56fe\u8868\u7c7b\u578b\u3001\u6570\u636e\u7b49\u4fe1\u606f\uff0c\u5e76\u6839\u636e\u6570\u636e\u96c6\u4e2d\u6ce8\u91ca\u7684\u5143\u4fe1\u606f\u8bc4\u4f30\u56fe\u8868\u7684\u5951\u5408\u6027\u3002\u8fd9\u4e00\u6b65\u662f\u4e3a\u4e86\u786e\u4fdd\u751f\u6210\u7684\u56fe\u8868\u80fd\u591f\u7b26\u5408\u7528\u6237\u67e5\u8be2\u7684\u8981\u6c42\uff0c\u5e76\u4e14\u6570\u636e\u6620\u5c04\u5408\u7406\u3002 \u53ef\u8bfb\u6027\u8bc4\u4f30\u662f\u6846\u67b6\u4e2d\u6700\u5177\u6311\u6218\u6027\u7684\u90e8\u5206\u3002\u8bc4\u4f30\u53ef\u8bfb\u6027\u9700\u8981\u8003\u8651\u591a\u65b9\u9762\u56e0\u7d20\uff0c\u4f8b\u5982\u5e03\u5c40\u3001\u6bd4\u4f8b\u548c\u989c\u8272\u7b49\u3002\u4e3a\u6b64\uff0c\u7814\u7a76\u5458\u4eec\u501f\u52a9\u4e86 GPT-4V(ISION) \u7684\u5f3a\u5927\u80fd\u529b\uff0c\u5e76\u5c06\u53ef\u8bfb\u6027\u8bc4\u4f30\u4efb\u52a1\u5206\u89e3\u4e3a\u591a\u4e2a\u66f4\u53ef\u63a7\u7684\u5b50\u95ee\u9898\u3002\u5148\u5206\u522b\u5bf9\u5e03\u5c40\u3001\u6bd4\u4f8b\u5c3a\u548c\u523b\u5ea6\u8fdb\u884c\u7ec6\u5316\u8bc4\u4f30\uff0c\u518d\u5c06\u8bc4\u4f30\u7ed3\u679c\u6c47\u603b\uff0c\u6574\u4f53\u7ed9\u51fa\u53ef\u8bfb\u6027\u7684\u6253\u5206\u3002\u7814\u7a76\u5458\u4eec\u7684\u5b9a\u91cf\u5b9e\u9a8c\u8868\u660e\uff0cVisEval \u7684\u53ef\u8bfb\u6027\u8bc4\u4f30\u7ed3\u679c\u4e0e\u4eba\u7c7b\u504f\u597d\u9ad8\u5ea6\u4e00\u81f4\u3002 \u57fa\u4e8e VisEval \u6846\u67b6\uff0c\u7814\u7a76\u5458\u4eec\u5bf9\u591a\u79cd\u6a21\u578b\u7684\u53ef\u89c6\u5316\u751f\u6210\u80fd\u529b\u8fdb\u884c\u4e86\u7cfb\u7edf\u8bc4\u4f30\u3002 \u7b2c\u4e00\u9879\u5b9e\u9a8c\u6d4b\u8bd5\u4e86\u4e0d\u540c\u6a21\u578b\u5728\u4f7f\u7528 Matplotlib \u548c Seaborn \u5e93\u65f6\u7684\u8868\u73b0\u3002\u5982\u88681\u6240\u793a\uff0cVisEval \u63ed\u793a\u4e86\u4e0d\u540c\u6a21\u578b\u5728\u751f\u6210\u53ef\u89c6\u5316\u65b9\u9762\u7684\u663e\u8457\u5dee\u5f02\u3002\u4ee5 GPT-4 […]<\/p>\n","protected":false},"author":34512,"featured_media":1093782,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-content-parent":1012650,"footnotes":""},"research-area":[],"msr-locale":[268881],"msr-post-option":[],"class_list":["post-1093740","msr-blog-post","type-msr-blog-post","status-publish","has-post-thumbnail","hentry","msr-locale-zh_cn"],"msr_assoc_parent":{"id":1012650,"type":"lab"},"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-blog-post\/1093740","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-blog-post"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-blog-post"}],"author":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/users\/34512"}],"version-history":[{"count":6,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-blog-post\/1093740\/revisions"}],"predecessor-version":[{"id":1093785,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-blog-post\/1093740\/revisions\/1093785"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media\/1093782"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1093740"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1093740"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1093740"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1093740"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}<\/figure>\n\n\n\n
\u6570\u636e\u53ef\u89c6\u5316\u7684\u667a\u80fd\u8fb9\u754c<\/h2>\n\n\n\n
\n
VisEval\uff1a\u9ad8\u8d28\u91cf\u6570\u636e\u96c6\u4e0e\u81ea\u52a8\u5316\u8bc4\u4f30\u6846\u67b6<\/h2>\n\n\n\n
\n
\n
\n
\u5b9e\u9a8c\u7ed3\u679c<\/h2>\n\n\n\n